Registro fonte: Special session: how much quality is enough quality? A case for acceptability in approximate designs

Special session: how much quality is enough quality? A case for acceptability in approximate designs

Approximate systems are designed to offer improved efficiency with potentially reduced quality of results. Quality of output in these systems is typically quantified in comparison to a precise result using metrics such as RMSE, MAE, PSNR, or application-specific metrics such as structural similarity...

ver descrição completa

Principais autores:	Felzmann, Isaías Bittencourt, Fabrício Filho, João, Oliveira, Juliane Regina de, Wanner, Lucas Francisco
Formato:	Trabalho Apresentado em Evento
Idioma:	Inglês
Publicado em:	Campo Mourao 2022
Assuntos:	Teoria da aproximação Imagens digitais Processamento de sinais Energia - Consumo Approximation theory Digital images Signal processing Energy consumption CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
Acesso em linha:	http://repositorio.utfpr.edu.br/jspui/handle/1/29755
Tags:	Adicionar Tag Sem tags, seja o primeiro a adicionar uma tag!

id	riut-1-29755
recordtype	dspace
spelling	riut-1-297552022-09-28T06:07:13Z Special session: how much quality is enough quality? A case for acceptability in approximate designs Felzmann, Isaías Bittencourt Fabrício Filho, João Oliveira, Juliane Regina de Wanner, Lucas Francisco Teoria da aproximação Imagens digitais Processamento de sinais Energia - Consumo Approximation theory Digital images Signal processing Energy consumption CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO Approximate systems are designed to offer improved efficiency with potentially reduced quality of results. Quality of output in these systems is typically quantified in comparison to a precise result using metrics such as RMSE, MAE, PSNR, or application-specific metrics such as structural similarity of images (SSIM). Furthermore, systems are typically designed to maximize efficiency for a given minimum quality requirement. It is often difficult to determine what this quality requirement should be for an application, let alone a system. Thus, a fixed quality requirement may be overly conservative, and leave optimization opportunities on the table. In this work, we present a different approach to evaluate approximate systems based on the usefulness of results instead of quality. Our method qualitatively determines the acceptability of approximate results within different processing pipelines. To demonstrate the method, we implement three image and signal processing applications featuring scenarios of image classification, image recognition, and frequency estimation. Our results show that designing approximate systems to guarantee acceptability can produce up to 20% more valid results than the conservative quality thresholds commonly adopted in the literature, allowing for higher error rates and, consequently, lower energy cost. 2022-09-27T17:01:34Z 5000 2022-09-27T17:01:34Z 2021-10-24 conferenceObject FELZMANN, Isaías; FABRÍCIO FILHO, João; OLIVEIRA, Juliane Regina de; WANNER, Lucas. Special session: how much quality is enough quality? A case for acceptability in approximate designs. In: INTERNATIONAL CONFERENCE ON COMPUTER DESIGN, 39., 2021, Storrs. Anais eletrônicos […]. Los Alamitos: IEEE Computer Society, Conference Publishing Services, 2021. p. 5-8. DOI: 10.1109/ICCD53106.2021.00013. Disponível em: https://ieeexplore.ieee.org/document/9643638. Acesso em: 09 jun. 2022. 9781665432191 2576-6996 http://repositorio.utfpr.edu.br/jspui/handle/1/29755 10.1109/ICCD53106.2021.00013 eng International Conference on Computer Design https://ieeexplore.ieee.org/document/9643638 embargoedAccess https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9643822 application/pdf Campo Mourao Estados unidos
institution	Universidade Tecnológica Federal do Paraná
collection	RIUT
language	Inglês
topic	Teoria da aproximação Imagens digitais Processamento de sinais Energia - Consumo Approximation theory Digital images Signal processing Energy consumption CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
spellingShingle	Teoria da aproximação Imagens digitais Processamento de sinais Energia - Consumo Approximation theory Digital images Signal processing Energy consumption CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO Felzmann, Isaías Bittencourt Fabrício Filho, João Oliveira, Juliane Regina de Wanner, Lucas Francisco Special session: how much quality is enough quality? A case for acceptability in approximate designs
description	Approximate systems are designed to offer improved efficiency with potentially reduced quality of results. Quality of output in these systems is typically quantified in comparison to a precise result using metrics such as RMSE, MAE, PSNR, or application-specific metrics such as structural similarity of images (SSIM). Furthermore, systems are typically designed to maximize efficiency for a given minimum quality requirement. It is often difficult to determine what this quality requirement should be for an application, let alone a system. Thus, a fixed quality requirement may be overly conservative, and leave optimization opportunities on the table. In this work, we present a different approach to evaluate approximate systems based on the usefulness of results instead of quality. Our method qualitatively determines the acceptability of approximate results within different processing pipelines. To demonstrate the method, we implement three image and signal processing applications featuring scenarios of image classification, image recognition, and frequency estimation. Our results show that designing approximate systems to guarantee acceptability can produce up to 20% more valid results than the conservative quality thresholds commonly adopted in the literature, allowing for higher error rates and, consequently, lower energy cost.
format	Trabalho Apresentado em Evento
author	Felzmann, Isaías Bittencourt Fabrício Filho, João Oliveira, Juliane Regina de Wanner, Lucas Francisco
author_sort	Felzmann, Isaías Bittencourt
title	Special session: how much quality is enough quality? A case for acceptability in approximate designs
title_short	Special session: how much quality is enough quality? A case for acceptability in approximate designs
title_full	Special session: how much quality is enough quality? A case for acceptability in approximate designs
title_fullStr	Special session: how much quality is enough quality? A case for acceptability in approximate designs
title_full_unstemmed	Special session: how much quality is enough quality? A case for acceptability in approximate designs
title_sort	special session: how much quality is enough quality? a case for acceptability in approximate designs
publisher	Campo Mourao
publishDate	2022
citation	FELZMANN, Isaías; FABRÍCIO FILHO, João; OLIVEIRA, Juliane Regina de; WANNER, Lucas. Special session: how much quality is enough quality? A case for acceptability in approximate designs. In: INTERNATIONAL CONFERENCE ON COMPUTER DESIGN, 39., 2021, Storrs. Anais eletrônicos […]. Los Alamitos: IEEE Computer Society, Conference Publishing Services, 2021. p. 5-8. DOI: 10.1109/ICCD53106.2021.00013. Disponível em: https://ieeexplore.ieee.org/document/9643638. Acesso em: 09 jun. 2022. 9781665432191 2576-6996 10.1109/ICCD53106.2021.00013
url	http://repositorio.utfpr.edu.br/jspui/handle/1/29755
_version_	1805318189078282240
score	10,814766

Special session: how much quality is enough quality? A case for acceptability in approximate designs

Registros relacionados